Fingerprinting Localization in Sigfox Networks: Analysis of the Impact of Base Stations with a Small Number of Received Messages

Le résumé fourni par la source

The adoption of Sigfox technology into Internet of Things (IoT) applications is on the rise thanks to its low cost, low power, and long-range wireless communication. Estimating locations of target nodes in Sigfox networks can be done by adopting range or fingerprinting-based localization approaches, with the latter increasingly gaining attention from researchers because of its robustness and effectiveness in multipath and non-line-of-sight (NLOS) signal propagation scenarios. Since fingerprinting-based localization in Sigfox networks is still a relatively new field of research, more effort is needed from research communities to improve the performance of existing methods. A well-structured Sigfox dataset is key to achieving high localization accuracies using fingerprinting localization models. With this in mind, this work attempts to determine the impact of the base stations with fewer received gateway messages on the localization accuracy of fingerprinting-based methods. The findings from this study indicate that a sparse dataset, due to the inclusion of many base stations with fewer received messages, could lower the accuracy of deep learning-based localization methods. Their exclusion could improve overall localization performance. However, it should not be excessive to avoid depriving the models of the ability to extract useful features necessary to infer locations of target nodes with relatively high accuracies.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.